Andrzej Rzonca of the Civic Development Forum and Polish central bank's Monetary Policy Council presented his paper "The consequences of unconventional monetary policy: what central banks do not take into account in their models?" at the 131st mBank - CASE seminar. What promised to be an insightful look at the shortcomings of central banks turned out to be a painful example of a bad modelling strategy.
Rzonca started his presentation by comparing the macroeconomic indicators in the United States and Europe, using debt data from both (?) the IMF's World Economic Outlook at Reinhart and Rogoff (Yes, the two Harvard professors who frauded their right-wing iconic paper on national debt and growth). He then claimed that central banks used neo-keynesian models and stated that central banks should consider 4 phenomena that impact the economy: restructuring, uncertainty, credit and public debt.
Having claimed that central banks use neo-keynesian models, it is quite surprising that the IS-LM curves, central to neo-keynesianism and representing investment (and in great part credit) and money demand, aren't considered by the author as modelling credit in the economy.
Putting the credit issue aside, high public debt was presented as a factor reducing growth, which is in line with the Reinhart and Rogoff philosophy yet is inconsistent with empirical findings. As a case in point, Japan's high debt amounting to 244% of GDP is not described as the cause for slow growth by any serious economic commentator.
Uncertainty surprisingly made its way on the list of grievances as if central banks had never considered the idea that the future might not look the way they imagine it.
Numerical models in macroeconomics are created in a way that allows predictions to be made. However, the more variables you input in the model, the harder it is to construct it, use it, track it, and ultimately understand it. Additionally, for statistical reasons, you want to keep the number of inputs the lowest possible without omitting any important one. Furthermore, not everything can be inserted in a model: that which cannot be measured just can't find its way in a model. This is why uncertainty is not included in such models (and uncertainty is NOT the same as risk): it can't be measured (unless you believe uncertainty to exist when it is mentioned in the media, which is ludicrous). Finally, there a nearly infinite amount of phenomena which may have an impact on the economy - they are not included in models of the economy because they have no empirically significant effect on it.
It is quite revealing that the Rzonca paper focused on theoretical causal links rather than empirical ones. What we have at play, with Rzonca's paper, is not sound and novel macroeconomic thinking but rather the recycled European austerity discourse linking public debt to the confidence fairy which is presented as uncertainty (confidence in the market, defended by austerians in Europe, was never demonstrated empirically). The austerian discourse is a bid to slash public deficits by reducing governmental redbuiistribution efforts. Although the confidence fairy has never been demonstrated to exist, empirical research rather suggests that non-extreme redistribution of wealth favors economic growth (see the IMF study Redistribution, Inequality, and Growth).
Central banks certainly do use neo-keynesian models, but they mainly use DSGE (Dynamic Stochastic General Equilibrium) models. Perhaps what Rzonca really meant to say (or did say - my Polish isn't that good yet) was that central banks should include financial frictions, transmission channels and risks in their models: in that case, he should have presented the Macrofinancial Modeling at Central Banks: Recent Developments and Future Directions IMF working paper that does exactly that, which also happened to describe how central banks had actually started doing it in 2010, 4 years ago. Andrzej Rzonca has either poorly attempted to discuss central bank modelling without doing basic literature review or poorly attempted to copy established publications in the field. Either way, it's not pretty.
Rzonca started his presentation by comparing the macroeconomic indicators in the United States and Europe, using debt data from both (?) the IMF's World Economic Outlook at Reinhart and Rogoff (Yes, the two Harvard professors who frauded their right-wing iconic paper on national debt and growth). He then claimed that central banks used neo-keynesian models and stated that central banks should consider 4 phenomena that impact the economy: restructuring, uncertainty, credit and public debt.
Having claimed that central banks use neo-keynesian models, it is quite surprising that the IS-LM curves, central to neo-keynesianism and representing investment (and in great part credit) and money demand, aren't considered by the author as modelling credit in the economy.
Putting the credit issue aside, high public debt was presented as a factor reducing growth, which is in line with the Reinhart and Rogoff philosophy yet is inconsistent with empirical findings. As a case in point, Japan's high debt amounting to 244% of GDP is not described as the cause for slow growth by any serious economic commentator.
Uncertainty surprisingly made its way on the list of grievances as if central banks had never considered the idea that the future might not look the way they imagine it.
Numerical models in macroeconomics are created in a way that allows predictions to be made. However, the more variables you input in the model, the harder it is to construct it, use it, track it, and ultimately understand it. Additionally, for statistical reasons, you want to keep the number of inputs the lowest possible without omitting any important one. Furthermore, not everything can be inserted in a model: that which cannot be measured just can't find its way in a model. This is why uncertainty is not included in such models (and uncertainty is NOT the same as risk): it can't be measured (unless you believe uncertainty to exist when it is mentioned in the media, which is ludicrous). Finally, there a nearly infinite amount of phenomena which may have an impact on the economy - they are not included in models of the economy because they have no empirically significant effect on it.
It is quite revealing that the Rzonca paper focused on theoretical causal links rather than empirical ones. What we have at play, with Rzonca's paper, is not sound and novel macroeconomic thinking but rather the recycled European austerity discourse linking public debt to the confidence fairy which is presented as uncertainty (confidence in the market, defended by austerians in Europe, was never demonstrated empirically). The austerian discourse is a bid to slash public deficits by reducing governmental redbuiistribution efforts. Although the confidence fairy has never been demonstrated to exist, empirical research rather suggests that non-extreme redistribution of wealth favors economic growth (see the IMF study Redistribution, Inequality, and Growth).
Central banks certainly do use neo-keynesian models, but they mainly use DSGE (Dynamic Stochastic General Equilibrium) models. Perhaps what Rzonca really meant to say (or did say - my Polish isn't that good yet) was that central banks should include financial frictions, transmission channels and risks in their models: in that case, he should have presented the Macrofinancial Modeling at Central Banks: Recent Developments and Future Directions IMF working paper that does exactly that, which also happened to describe how central banks had actually started doing it in 2010, 4 years ago. Andrzej Rzonca has either poorly attempted to discuss central bank modelling without doing basic literature review or poorly attempted to copy established publications in the field. Either way, it's not pretty.
No comments:
Post a Comment